[pymvpa] BoxcarMapper

[e c]

Hi guys,

I'm trying to use BoxcarMapper to do block averaging of 
my fmri data. I have blocks of length 8 TRs and want to average to one 
signal value per block. I manage to do this with transform_with_boxcar, 
but this only outputs the sample values without preserving the rest of 
the mvpa Dataset structure, so I'm now trying to use the mapper instead.

First I create my mapper: bcm =BoxcarMapper([0,8...etc],8)

I try to do: bcm(dataset) but I get the error:
<BoxcarMapper: bl=8> needs to be trained before it can be used and auto training is disabled.

What training does that refer to? Shouldn't the mapper just contain function that does simple averaging over TRs?

Thanks!
-Edmund Chong

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20110111/37cfa408/attachment.htm>